Validation of the Seizure-Related Impact Assessment Scale

Abstract

Background and Objectives

There is a clear need in epilepsy clinical trials and practice for a measure that captures the trade-off between seizure and treatment-related adverse effects, which is reliable over time and across different treatment regimens. We aimed to create and validate the Seizure-Related Impact Assessment Scale (SERIAS) to fill this need.

Methods

This was a prospective longitudinal study of adults with epilepsy recruited from an Australian comprehensive epilepsy center. Participants completed SERIAS at baseline and 3 and 6 months later. SERIAS has 6 self-report items. Five items record the number of days per month that seizures or treatment-related adverse effects partially or fully affect work/home/school and family/social/nonwork activities. The final item is an epilepsy disability visual analog scale. SERIAS is scored by adding the days per month of disability, with scores ranging from 0 to 150 (higher scores indicate more disability). SERIAS was completed alongside 7 validated instruments measuring seizure-related and treatment-related adverse effects (Work and Social Adjustment Scale [WSAS], Liverpool Adverse Events Profile [LAEP]), mood disorders (Neurological Disorders Depression Inventory for Epilepsy [NDDI-E], Generalized Anxiety Disorder [GAD-7]), somatic symptoms (Somatic Symptom Scale [SSS-8]), and quality of life (Quality of Life in Epilepsy Inventory [QOLIE]-31, EuroQol 5 Dimensions [EQ-5D]). General linear mixed models were used to investigate the relationship between the SERIAS and other relevant clinical and psychometric data. Standardized model coefficients β are presented with 95% confidence intervals.

Results

A total of 90 patients (64.4% female, mean age 43.1 years) completed baseline SERIAS. Most patients reported at least 1 day of disability (62%, median SERIAS score = 3, interquartile range = 18.3). Greater disability was negatively correlated with QOLIE-31 total score (β = −0.17, 95% CI −0.27 to −0.07) and positively correlated with scores on 5-level EQ-5D (β = 0.15, 95% CI 0.04–0.25), NDDI-E (β = 0.22, 95% CI 0.13–0.31), GAD-7 (β = 0.21, 95% CI 0.09–0.32), SSS8 (β = 0.29, 95% CI 0.17–0.41), LAEP (β = 0.29, 95% CI 0.20–0.39), WSAS seizure-related adverse events (β = 0.23, 95% CI 0.14–0.33), and WSAS treatment-related adverse events (β = 0.36, 95% CI 0.26–0.46). Higher seizure frequency was associated with higher SERIAS score (β = 0.07, 95% CI 0.03–0.11). Psychometric reliability for the SERIAS was acceptable (all coefficients >0.70) as was test-retest reliability (n = 35 patients, intraclass correlation coefficient = 0.72, 95% CI 0.51–0.85).

Discussion

SERIAS shows good psychometric reliability and strong test-retest stability. These findings suggest that SERIAS is a valid scale to measure epilepsy-related disability.

Introduction

Epilepsy exerts an immense burden of disease, affecting approximately 50 million people worldwide.^1-3 Two in 3 people living with epilepsy may become seizure free with antiseizure medications, and resective epilepsy surgery and neuromodulation may improve seizure control for many others.^4,5 Concerningly, epilepsy treatments may cause substantial adverse effects.⁶ These adverse effects may occur before treatments reach therapeutic levels, may have nonlinear relationships with treatment doses, and may differ greatly between individuals. Accordingly, international guidelines recommend that epilepsy treatment-related adverse effects are assessed regularly.⁷ In some cases, a treatment's adverse effects will clearly outweigh its benefits and prompt a change in therapy. However, in many cases, treatments will exert continual, mild-moderate adverse effects that need to be weighed against seizure frequency and severity. For example, a person's antiseizure medication may cause dizziness that limits some daily activity, but prevents their focal seizures from evolving into bilateral tonic-clonic events. They may add in a new treatment that renders them seizure free, but they need to adjust their work load because of treatment-related tiredness. How does the impact of these 2 treatment regimens compare? There is a clear need in epilepsy clinical trials and practice for a measure that captures epilepsy disability, attributable to both seizure and treatment-related adverse effects, and the trade-off between the 2 that is reliable over time and across different treatment regimens. Disability refers to the reduced ability or inability to carry out usual activities, such as work, school, or household duties, because of disease-related impairment.⁸ Examples of epilepsy-related disability include taking extended time off work because of prolonged post-ictal confusion (seizure-related effect) or being unable to participate in social activities because of extreme fatigue (medication-related effect).

Patient-reported outcome measures (PROMs) offer a potential solution. These are usually brief, easy-to-understand instruments completed by patients to capture a range of important outcomes. Several epilepsy PROMs already capture the severity and disability caused by seizures and treatment-related adverse effects. These include the Quality of Life in Epilepsy Inventory (QOLIE) scales that measure health-related quality of life⁹ and Liverpool Adverse Events Profile (LAEP) that measures antiseizure medication-related adverse effects.¹⁰ However, they have several important limitations. For example, QOLIE-31 asks respondents to mark how bothersome their seizures are on a Likert scale, from “not at all bothersome” through to “extremely bothersome.” This does not provide granular detail on seizure severity, which may range from brief, focal aware seizures to prolonged bilateral tonic-clonic seizures, nor does it capture seizure frequency. Both these data points are needed to assess whether and how treatments are improving seizure burden. LAEP provides granular data on the type and severity of antiseizure medication-related adverse effects, but there is no sense of how much these interfere with people's ability to function. Finally, it is difficult to correlate the “trade-off” between seizure-related and treatment-related burden across 2 separate instruments and to “translate” and compare episodic seizure events against constant treatment-related adverse effects.

In this prospective, longitudinal study, we aimed to create and validate the Seizure-Related Impact Assessment Scale (SERIAS), a PROM that captures the burden related to seizure vs treatment adverse effects and provides an overall sense of epilepsy disability.

Methods

A detailed study protocol has been published;¹¹ this section describes the study methodology.

Setting and Participants

This prospective, longitudinal study was conducted at Alfred Health, an adult comprehensive epilepsy center in Melbourne, Australia. Participants were recruited from epilepsy inpatient and outpatient settings. Inclusion criteria were people age 18 years and older with an epileptologist-confirmed epilepsy diagnosis. Exclusion criteria were functional seizures and inability to complete English language-validated questionnaires. Concomitant anxiety, depression, and personality disorders were not exclusion criteria. Given that these are highly prevalent disorders in people with epilepsy, it was felt important to include and adjust for these to ensure SERIAS is validated in a cohort that is generalizable to as many people living with epilepsy as possible. Participants provided verbal and written informed consent.

Questionnaires

Participants completed 2 versions of SERIAS and a set of widely used, validated PROMs at baseline and 3 and 6 months later.

Text Version SERIAS

This 6-item self-report questionnaire has 5 items requesting participants to recall the number of days in the last month (30 days) that seizures or treatment-related adverse effects partially or fully impact work, home, school, and family/social/nonwork activities. The total score is computed as number of days affected summed across all activity domains (sum of SERIAS questions 1–5). Two subscale scores were computed to summarize the impact of seizures (SERIAS questions 1, 3, and 4) and treatment (SERIAS questions 2 and 5). The sixth item is a visual analog scale, requesting participants to circle a number between 0 (no impact) and 10 (severe impact) to indicate the overall impact of seizures-related and treatment-related adverse effects on their lives. As a worked example, a person may have experienced 1 seizure in the 30-day recall period, and this prevented them from working (question 1 = 1 point) and attending a social function (question 4 = 1 point) that day. They were otherwise well. Their total SERIAS score is 2.

This version of SERIAS was created by J.F., a senior epileptologist with extensive experience leading multicenter international clinical trials and was further refined with feedback from the study team including a lived experience researcher (C.E.). The format of the SERIAS was informed by the Migraine Disability Assessment Scale (MIDAS).¹² MIDAS is a 5-item instrument that asks participants to recall the number of days over the previous 3 months where migraine partially or fully impacted school or work, household work or chores, and family/social/leisure activities. It is widely used in clinical migraine research and practice and is one of the most implemented condition-specific PROM in new neurologic drug approvals in Europe.^12,13

This text version of SERIAS is presented in Figure 1.

Figure 1. Text Version SERIAS.

Adapted from Foster E, Conquest A, Ewart C, et al. Validation of the Seizure-Related Impact Assessment Scale (SERIAS): a study protocol. BMJ Open. 2024;14(6):e083929. doi:10.1136/bmjopen-2024-083929.

Traffic Light SERIAS

An alternative, 3-item self-report questionnaire provided a pictorial prompt for capturing these data. The first 2 items ask participants to recall the number of days in the last month (30 days) that were “green light” (no major symptoms), “yellow light” (somewhat affected), or “red light” (greatly affected) for seizure-related (question 1) and treatment-related adverse effects (question 2). The total score is computed as the total number of “red light” days multiplied by 2 and added to the total number of “yellow light” days. The third item is the same visual analog scale included in the text version SERIAS.

The traffic light version of SERIAS was created by senior epileptologists that have extensive experience leading multicenter international clinical trials (P.K., J.F.), a neurologist (E.F.) with expertise in epilepsy and headache (both episodic conditions), a clinical neuropsychologist and biostatistician experienced in instrument validation (C.M.), and an experienced qualitative researcher (A.C.).^14,15 A lived experience researcher (C.E.) provided valuable feedback on the content, wording, and lay-out of SERIAS.

Figure 2 presents the “traffic light” version of SERIAS.

Figure 2. Traffic Light SERIAS.

Adapted from Foster E, Conquest A, Ewart C, et al. Validation of the Seizure-Related Impact Assessment Scale (SERIAS): a study protocol. BMJ Open. 2024;14(6):e083929. doi:10.1136/bmjopen-2024-083929.

Participants were asked to complete both versions of SERIAS at each time point and to nominate which they preferred and why.

In addition to the 2 versions of SERIAS, participants answered “what date was your last seizure” and completed LAEP (antiseizure medication-related adverse effects),¹⁰ Somatic Symptom Scale (SSS-8, generic somatic symptoms),¹⁶ Work and Social Adjustment Scale (WSAS, repeated for both seizure- and treatment-related adverse effects),¹⁷ QOLIE-31,⁹ EQ-5D (generic quality of life),¹⁸ Neurological Disorders Depression Inventory for Epilepsy (NDDI-E),¹⁹ and Generalized Anxiety Disorder (GAD-7).²⁰ In the absence of appropriate validated instruments, researchers custom-built brief questionnaires for eligible participants regarding adverse effects relating to neuromodulation, ketogenic diet, and epilepsy surgery. These were completed alongside the SERIAS and validated instruments and are included in eAppendix 1.

At the 3-month and 6-month follow-up time points, participants were also asked: “Overall, compared with the last time you did these questionnaires, do you think the level of epilepsy-related impact (driven by seizures and the treatment used for seizures) has (1) meaningfully changed for the better, (2) changed for the worst, (3) stayed the same, or (4) unsure. Please pick one option.”

We prospectively collected relevant clinicodemographic data at baseline (e.g., age, sex, epilepsy type) and longitudinally (e.g., seizure type and frequency, antiseizure medication type and doses).

Study questionnaires and data were distributed, collected, and managed via Alfred Health's Research Electronic Data Capture (REDCap). REDCap is a secure, web-based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages; and (4) procedures for data integration and interoperability with external sources.^21,22 Paper-based questionnaires were made available to participants on request, and data were entered by researchers into the study REDCap.

Test-Retest Reliability Subgroup

Researchers also recruited a “test-retest” subgroup to test whether SERIAS scores remained stable within a tight timeframe (2-week period) during which time participants' seizures and treatment adverse effects were unlikely to substantially change. A consecutive group of participants were invited to complete SERIAS, NDDI-E, GAD7, and answered “what date was your last seizure?” 2 weeks before the baseline time point and then completed all instruments at baseline and 3 and 6 months, in line with the main study group.

Statistical Analysis

Before analysis, all key variables were inspected for outliers, missing data, skew, and kurtosis. For the analysis of convergent validity, separate general linear mixed models (GLMMs) were computed to assess the ability of the SERIAS total score to predict scores on the 10-point visual acuity scale (SERIAS question 6) and established and related instruments. For each model, a random intercept for participant was specified to allow data from all time points to be used. Similar models were estimated to investigate the relationships between SERIAS scores and clinical variables, including number of antiseizure medications and seizure frequency. Psychometric reliability was examined using 2 commonly used reliability coefficients: McDonald ω and Cronbach α.^23,24 Test-retest reliability was examined using Spearman r and Kendall tau correlation coefficients, and intraclass coefficient was also calculated. A GLMM with time as a predictor was computed to investigate longitudinal change in the SERIAS total score. SERIAS scores were log transformed for all analyses in which the SERIAS scores were entered as dependent variables. All continuous variables were scaled and centered before analysis. Standardized regression coefficients (β) are presented along with robust 95% bias-corrected and accelerated CIs computed via bootstrapping with 2,000 replicates. All analyses were performed using the R software (version 4.3.2).

Standard Protocol Approvals, Registrations, and Patient Consents

Ethics approval was granted by the Alfred Health Ethics Committee (HREC 17/23). The study is registered with the Australian Clinical Trials Registry (ACTRN 12623000599673).

Data Availability

Nonidentifiable data will be made available on request to appropriately qualified investigators, providing all necessary regulatory approvals are in place.

Results

Sample Characteristics

In total, 268 SERIAS questionnaires were completed by 105 participants at any time point. This included 90 participants at baseline, 70 participants at 3 months, and 62 participants at 6 months. Thirty-five participants completed the SERIAS 2 weeks before and at the baseline time point to generate test-retest data, as described below. At baseline, most participants were female (n = 58, 64%) with a mean age of 43.1 (SD = 14.6) years. The most common epilepsy diagnosis was focal epilepsy (n = 71, 79%), followed by generalized epilepsy (n = 16, 18%) and epilepsy with unknown onset (n = 3, 3%). At baseline, participants were on the following treatments: antiseizure medication monotherapy (n = 30, 33.3%), antiseizure medication polytherapy (n = 54, 60%), vagus nerve stimulation (n = 6, 6.7%), and previous epilepsy surgery (n = 19, 21.1%). At baseline, 52 (58%) patients had been seizure free for at least 12 months. The mean seizure score, based on an established seizure frequency scale, was “7.”²⁵ This correlates to a seizure frequency of 1–3 seizures per month.

SERIAS Seizure-Related Impact

Across all time points, the median SERIAS score was 2 (interquartile range [IQR] = 15, min = 0, max = 150), representing 2 days of seizure-related impact in the previous month. The distribution was highly skewed, with most participants indicating no or little impact. Overall, 119 (44%) participants who completed SERIAS questionnaires indicated no seizure-related impact, whereas the remaining 56% indicated at least 1 day of seizure-related impact over the 30-day recall period. SERIAS scores were classified into no impact (SERIAS total score = 0 days), low impact (SERIAS total score >0 and ≤7 days), moderate impact (SERIAS total score >7 and ≤25 days), and high impact (SERIAS total score >25 days). These results are shown in Table 1.

Table 1.

Displays Seizure-Related Impact at Each Time Point, as Measured by SERIAS Items 1, 3, and 4

	Total, median (IQR)	Level of impact, n (%)
		None	Low	Moderate	High
All time points (n = 268)	2 (15)	119 (44)	54 (20)	49 (18)	46 (17)
Pretest (n = 46)	1 (12.50)	22 (48)	10 (22)	5 (11)	9 (20)
Baseline (n = 90)	3 (18.25)	34 (38)	18 (20)	19 (21)	19 (21)
3 mo (n = 70)	1 (11.75)	34 (49)	13 (19)	14 (20)	9 (13)
6 mo (n = 62)	1 (11.50)	29 (47)	13 (21)	11 (18)	9 (15)

Abbreviations: IQR = interquartile range; SERIAS = Seizure-Related Impact Assessment Scale.

Item 6, the visual analog scale (0 = not affected, 10 = severely affected), revealed a median score of 1 (IQR = 5, min = 0, max = 10), with the majority of responses indicating some degree of impact (score >0, n = 170, 63%).

Convergent Validity With Other Psychometric Instruments

Higher scores on the SERIAS were associated with higher scores on the 10-point visual analog scale (Item 6) (β = 0.47, 95% CI 0.39–0.54), lower scores on the QOLIE-31 (β = −0.17, 95% CI −0.27 to −0.07), higher scores on the EQ-5D-5L (β = 0.15, 95% CI 0.04–0.25), NDDI-E (β = 0.22, 95% CI 0.13–0.31), higher scores on the GAD-7 (β = 0.2, 95% CI 0.09–0.32), higher scores on the SSS8 (β = 0.29, 95% CI 0.17–0.41), higher scores on the LAEP (β = 0.29, 95% CI 0.20–0.39), higher scores on the WSAS treatment-related adverse effects scale (β = 0.23, 95% CI 0.14–0.33), and higher scores on the WSAS seizure-related adverse effects (β = 0.36, 95% CI 0.26–0.46).

We separately examined the relationship between the SERIAS seizure score and the WSAS seizure-related adverse effects score, as well as the SERIAS treatment score and the WSAS treatment-related adverse effects score. The SERIAS seizure score was a strong predictor of the WSAS seizure-related adverse effects score (β = 0.37, 95% CI 0.28–0.46). The SERIAS treatment score was a strong predictor of the WSAS treatment-related adverse effects score (β = 0.23, 95% CI 0.13–0.32).

Convergent Validity With Clinical Variables

To further examine convergent validity, separate GLMMs were computed to examine the relationship between SERIAS scores, seizure frequency, and number of antiseizure medications. Higher SERIAS total scores were associated with the frequency of any seizure (β = 0.07, 95% CI 0.03–0.11), focal impaired awareness seizures (β = 0.10, 95% CI 0.04–0.16), and bilateral tonic-clonic seizures (β = 0.14, 95% CI 0.06–0.22). The SERIAS total score was not significantly predicted by the frequency of focal aware seizures (β = 0.00, 95% CI −0.07 to 0.07) or other seizure type (β = −0.01, 95% CI −0.12 to 0.11).

The SERIAS seizure score was significantly associated with the frequency of any seizure (β = 0.07, 95% CI 0.04–0.11), focal impaired awareness seizures (β = 0.12, 95% CI 0.05–0.18), and bilateral tonic-clonic seizures (β = 0.12, 95% CI 0.04–0.19). The SERIAS seizure score was not significantly associated with the frequency of focal aware seizures (β = 0.01, 95% CI −0.05 to 0.08) or other seizure types (β = 0.01, 95% CI −0.09 to 0.11).

The SERIAS treatment score was significantly associated with the frequency of bilateral tonic-clonic seizures (β = 0.1, 95% CI 0.04–0.17), but not total seizure frequency (β = 0.03, 95% CI 0.00–0.06), focal aware seizures (β = 0.01, 95% CI −0.04 to 0.06), focal impaired awareness seizures (β = 0.04, 95% CI −0.01 to 0.09), or other seizures (β = −0.08, 95% CI −0.17 to 0.01).

The number of antiseizure medications was not associated with the SERIAS total score (β = 0.13, 95% CI −0.01 to 0.28), SERIAS seizure score (β = 0.12, 95% CI −0.01 to 0.27), or the SERIAS treatment score (β = 0.08, 95% CI −0.03 to 0.20).

Psychometric Reliability

Psychometric reliability was examined using 2 commonly used reliability coefficients. At all time points, the SERIAS demonstrated acceptable reliability using both coefficients (McDonald ω and Cronbach α > 0.7). These results are shown in Table 2.

Table 2.

Displays SERIAS Acceptable Psychometric Reliability

	McDonald ω	Cronbach α
Pretest (n = 46)	0.91	0.84
Baseline (n = 90)	0.87	0.83
3 mo (n = 70)	0.92	0.91
6 mo (n = 62)	0.81	0.77

Abbreviation: SERIAS = Seizure-Related Impact Assessment Scale.

Test-Retest Reliability

Thirty-five participants in the test-retest group had completed all data points at baseline and 2 weeks prior. Correlation coefficients between pretest and baseline were statistically significant (Spearman r = 0.60, 95% CI 0.33–0.78; Kendall tau = 0.54, 95% CI 0.33–0.76). The intraclass correlation coefficient (ICC) was also high (ICC = 0.72, 95% CI 0.51–0.85).

Trajectories of SERIAS Scores

Overall, there was no evidence for longitudinal change in SERIAS scores over time (β = −0.05, 95% CI −0.11 to 0.00). This is displayed in Figure 3.

Figure 3. Longitudinal Trajectory of SERIAS Scores.

General linear mixed model demonstrates stable longitudinal trajectory of SERIAS total scores over the study period. SERIAS = Seizure-Related Impact Assessment Scale.

Traffic Light Version

Of the participants who completed the text version SERIAS, 62 also completed the traffic light SERIAS on at least 1 occasion, leading to 115 complete traffic light SERIAS scores. These scores are summarized in Table 3. The traffic light total score was a significant predictor of SERIAS total score (β = 0.65, 95% CI 0.53–0.76) indicating strong agreement. The traffic light seizure score was a significant predictor of the SERIAS seizure score (β = 0.70, 95% CI 0.58–0.82). Similarly, the traffic light treatment score was a significant predictor of the SERIAS treatment score (β = 0.41, 95% CI 0.30–0.52).

Table 3.

Displays Total Scores From the SERIAS Traffic Light Version

	Seizures	Treatment	Total
Green	26.50 (15.00)	25.00 (21.00)	50.00 (32.00)
Yellow	1.00 (7.00)	1.50 (10.00)	5.00 (19.75)
Red	0.00 (3.00)	0.00 (2.00)	0.00 (5.00)

Abbreviation: SERIAS = Seizure-Related Impact Assessment Scale.

In total, 132 ratings were obtained regarding the preference for the traffic light vs text version SERIAS. The most common responses indicated preference for the text version (n = 66, 49%), followed by the traffic light version (n = 33, 25%). The remaining responses indicated no preference (n = 33, 25%). Figure 4 displays the relationship between the total scores of the 2 versions of the SERIAS.

Figure 4. Relationship of 2 SERIAS Versions.

The close relationship between the total scores of the 2 versions of SERIAS. SERIAS = Seizure-Related Impact Assessment Scale.

Discussion

This study validates a novel PROM, SERIAS, to quantify seizure-related and treatment-related disability in adult patients with epilepsy. It is brief, easy to administer, and has good psychometric reliability. That is, higher SERIAS scores indicating worse disability are associated with higher seizure burden and higher medication adverse effects measured via existing validated instruments. Furthermore, SERIAS computes epilepsy-related disability into something tangible, that is, time lost from work, study, and social activities. SERIAS addresses several crucial issues and fills important gaps in epilepsy PROMs, and may be a useful tool in epilepsy clinical trials and practice.

Seizures drive much of the burden of epilepsy, and accordingly, change in seizure frequency may be used as a marker of treatment success.^26,27 However, there are several important drawbacks to this approach. First, seizure frequency is traditionally measured via patient or caregiver completion of seizure diaries. Seizure diaries are frequently inaccurate, with patients documenting fewer than half their seizures on average because of impaired perception (unaware that seizures occurred), volition (impaired ability to document seizure), or memory (forgot to record seizure).^28,29 Second, “uncountable seizures,” for example brief absences that are not obvious to either patients nor caregivers, may cumulatively exert a substantial impact on people's lives. These are not easily or accurately captured through seizure diaries. Third, seizure frequency does not necessarily translate to disease severity; other critical factors include ictal features and likelihood of injury. Seizures may manifest with a wide range of ictal features (e.g., urinary incontinence, musculoskeletal injury). Although several PROMs specifically ask after objective features (e.g., seizure duration, loss of consciousness), they do not capture all possibilities (e.g., an impending sense of doom).³⁰ This means disabling seizure features may not be accounted for. Furthermore, the same feature will exert different disability depending on individual circumstances. For example, urinary incontinence may require a person working in an office to take more time off work as they return home to change, compared with someone working from home. SERIAS overcomes the shortcomings of “seizure counting” as it converts seizure frequency and ictal features into a common denominator, that is, “time lost.” This is a novel and robust approach to measuring seizure-related disability as the shared feature of all disabling seizures is the time they take away from people's daily activities.

The SERIAS score captures the “trade-off” between seizure control and treatment-related adverse effects, not readily measurable with current instruments. This is essential because treatment efficacy and tolerability are “two sides of the same coin,”³¹ and many people living with epilepsy would not trade off treatment-related adverse effects that cause daily dysfunction for modest or moderate seizure reduction. Most of the comparable PROMs focus on seizure-related or treatment-related adverse effects in isolation. For example, the Global Assessment of Severity of Epilepsy is a 7-point Likert scale ranging from “not at all severe” to “extremely severe,” but does not consider treatment-related adverse effects. Conversely, the LAEP measures treatment-related adverse effects, without considering seizure-related adverse effects.^10,32 Quality of Life in Epilepsy instruments (e.g., QOLIE-89, questions 65–68) ask patients to rate on a 5-point Likert scale how much epilepsy or antiseizure medication has caused³³ difficulty with working, relationships, leisure time, and driving (“not at all” through to “a great deal”), but does not separate the impact of seizures from medication adverse effects.⁹ In contrast to these instruments, QOLIE-31 and the Personal Impact of Epilepsy Scale (PIES) both ask patients to rate on Likert scales the impact of seizures and treatment separately.³⁴ However, these impacts are measured qualitatively, making it difficult to directly “trade off” seizure vs treatment-related adverse effects, and both instruments are substantially longer than SERIAS (31 items for QOLIE-31 and 25 items for PIES). Furthermore, PIES has a longer recall period (3 months) than SERIAS, which increases the risk of recall bias. SERIAS' separate consideration of seizure and treatment burden, converted into the same quantifiable end point (time lost) allows for a more objective and directly comparable assessment of the disability caused by these related but separate factors.

There is an ongoing need for simple, pragmatic PROMs in epilepsy clinical trials and practice. Our study demonstrates that SERIAS is “fit for purpose”; it correlates well with relevant PROMs that capture quality of life, disability, and treatment adverse effects. From a clinical trials perspective, SERIAS should be sensitive to reliably measure disability over time. This is essential for demonstrating a treatment's efficacy. From a clinical practice perspective, it addresses the issue of “uncountable seizures” and the fact that number of seizures does not necessarily translate to the seizure burden. All these factors must be weighed when considering change in treatment regimens, including workup for epilepsy surgery or neuromodulation devices. The brevity and straightforward nature of SERIAS means it would be practicable to administer before each clinical encounter. This parallels with MIDAS, which is now embedded in migraine clinical trials and real-world practice. MIDAS scores that indicate worsening migraine or high migraine disability prompt discussions with clinicians about changing treatment regimens. Furthermore, initiation and ongoing access to more advanced migraine treatment may be guided by regular MIDAS scores. SERIAS may be similarly useful for epilepsy.

Limitations

First, participants were recruited from a comprehensive epilepsy center and so results may not be generalizable to other settings. However, it should be noted that our study cohort had a spread of ages, epilepsy-related disability, and epilepsy duration. Second, the US Food and Drug Administration states that PROMs should measure outcomes that are meaningful to patients, and these may be identified through concept elicitation and cognitive interviews.³⁵ Although a lived experience researcher provided robust input regarding SERIAS' content, wording, and formatting, the underlying structure is based on a widely used PROM for another episodic neurologic condition (MIDAS for migraine). Third, SERIAS relies on a 30-day recall period and so may be affected if people do not recall events accurately. Shorter time periods may be recalled more accurately (e.g., EQ-5D self-reports items for that day), but may not be representative of the recent epoch. A 30-day recall period is a common epoch used in epilepsy PROMs (e.g., QOLIE instruments). Future studies could investigate whether SERIAS may be useful in shorter interval assessments. Fourth, SERIAS has been validated only in adult epilepsy populations, and additional studies will need to validate 1 or both version of SERIAS for use in pediatric patients with epilepsy. Fifth, both seizures and treatment adverse effects may cause similar symptoms (e.g., mood disturbance and cognitive fogging), and it may be difficult to determine whether seizures or treatment is driving the burden of disease in each case. This limitation is partially overcome by the visual analog scale included as the final item on the questionnaire, which provides an overall sense of how people are doing. Sixth, despite attempts to encourage participation at all time points, a number of participants were lost to follow-up and did not complete all time points specified in the study design. Specifically, although 90 participants completed the protocol at baseline, only 62 completed the protocol at 6 months (69% completion). Our use of GLMMs partially mitigated against the statistical implications of participant drop out by allowing all participants to contribute to the models even if they had missing data at one or more time points (as long as at least 1 datapoint was available). Nevertheless, it is possible that participant dropout could be conditional on disease-related variables, including seizure-related and treatment-related disability. One of the implications of this is that, if the SERIAS is used in clinical outcome studies, care must be taken to ensure complete data collection. If this is not possible, the reasons for participant drop out should be investigated to ensure that missing SERIAS scores are not associated with high disability. Seventh, in our sample, SERIAS scores were highly skewed. Although the range of possible scores was between 0 and 150, the median score was 2. This indicates that most participants reported no or little treatment-related or seizure-related disability. One reason for this is that our sample comprised patients who generally had well-controlled epilepsy. Specifically, 58% of our sample were seizure free at the time of enrollment. It is important to note however, that the positive skew was driven by a significant proportion of patients with high disability. For example, at baseline 21% of patients indicated >25 days of treatment-related or seizure-related impact. This suggests that, although the median disability of the sample was relatively low, the SERIAS is sensitive to patients experiencing the highest levels of disability. Nevertheless, it will be important to investigate the performance of the SERIAS in patient populations with higher seizure frequency and poorly controlled disease.

SERIAS is a brief and reliable PROM for measuring epilepsy-related disability because of seizures and treatment-related adverse effects. It uses a novel approach to overcome shortcomings in existing epilepsy PROMs and ultimately may replace some PROMs. One of its clear advantages is capturing the “see-saw” between seizure control and treatment adverse effects by converting the impact of both of these into the same measure, that is, “time lost” from daily activities. The balance between seizure-related vs treatment-related adverse effects may be explored in more detail in future studies. Future randomized controlled trials may be able to determine whether SERIAS is sensitive and specific to detect change in disability for people receiving treatment compared to the placebo group and to include consumer ranking of “clinical improvement” to establish clinically meaningful change criterion.

Glossary

EQ-5D-5L

5-level EuroQol 5 Dimensions

GAD-7

Generalized Anxiety Disorder Scale

GLMM

general linear mixed model

ICC

intraclass correlation coefficient

IQR

interquartile range

LAEP

Liverpool Adverse Events Profile

MIDAS

Migraine Disability Assessment Scale

NDDI-E

Neurological Disorders Depression Inventory for Epilepsy

PIES

Personal Impact of Epilepsy Scale

PROM

patient reported outcome measure

QOLIE

Quality of Life in Epilepsy Inventory

REDCap

Research Electronic Data Capture

SERIAS

Seizure-Related Impact Assessment Scale

SSS-8

Somatic Symptom Scale

WSAS

Work and Social Adjustment Scale

Author Contributions

E. Foster: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. A. Conquest: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. C. Ewart: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. J.-P. Nicolo: drafting/revision of the manuscript for content, including medical writing for content. G Rayner: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. T.T. Winton-Brown: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. T.J. O'Brien: drafting/revision of the manuscript for content, including medical writing for content. P. Kwan: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data. C. Malpas: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data. J. French: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data.

Study Funding

This work was supported by an investigator-initiated research grant from LivaNova USA. This funding supports fees related to ethics submission and publication costs, and salary support for E. Foster, A. Conquest, and C. Malpas.

Disclosure

E. Foster reports salary support from LivaNova USA for the SERIAS validation study; and she and her institution report grants from the Avant Research Foundation, Brain Foundation Australia, the Clive and Vera Ramaciotti Foundation (Ramaciotti Health Investment Grant), GPCE, LivaNova (Australia and USA), Lundbeck (Australia), Monash Partners STAR Clinician Fellowship, Monash University Early Career Postdoctoral Fellowship, Royal Australasian College of Physicians Fellows Research Establishment Fellowship, and the Sylvia and Charles Viertel Charitable Foundation Clinical Investigator Award, and UCB, outside the submitted work. A. Conquest reports salary support from LivaNova USA for the SERIAS validation study; and she and/or her institution declare research support from Brain Foundation (Australia), outside the submitted work. J.-P. Nicolo and/or his institution have received research support and/or consultancy fees from Eisai, outside the submitted work. T.J. O'Brien and/or his institution have received research support from Australian and US government research granting organisation (NHMRC Program Grant #APP1091593, MRFF, DOD, National Institute of Neurological Disorders and Stroke); and research and/or consultancy fees from Industry: Biogen, Eisai Pharma, Jazz Pharmaceuticals, LivaNova, ES Therapeutics, Kinosis Pharmaceuticals, Supernus, UCB Pharma, and Zynerba Pharmaceuticals, outside the submitted work. P. Kwan is supported by a NHMRC Investigator Grant (GNT2025849); and he and/or his institution have received research support or consultancy fees from Angelini, Eisai, Jazz Pharmaceuticals, LivaNova, SK Life Science, and UCB Pharma, outside the submitted work. C. Malpas reports salary support from LivaNova USA for the SERIAS validation study; has received conference travel support and/or speaker fees from Merck, Novartis, and Biogen, outside the submitted work; and has received research support from the National Health and Medical Research Council, Multiple Sclerosis Research Australia, the University of Melbourne, the Royal Melbourne Hospital Neuroscience Foundation, and Dementia Australia, outside the submitted work. J. French consults for LivaNova; is president of the Epilepsy Study Consortium that owns the copyright to SERIAS; receives salary support from the Epilepsy Foundation and from the Epilepsy Study Consortium for consulting work and/or attending scientific advisory boards for Acadia Pharmaceuticals, Access Industries, Acuta Capital Partners, AFASCI Inc., Agrithera Inc., Alterity Therapeutics Ltd., Angelini Pharma S.p.A, Autifony Therapeutics Ltd., Axonis Therapeutics, Baergic Bio, Beacon Biosignals Inc., Biogen, Biohaven Pharmaceuticals, Bloom Science Inc., Bright Minds Biosciences Inc., Camp4 Therapeutics Corporation, Capsida Biotherapeutics, Cerebral Therapeutics, Cerecin Inc., Cerevel, Ceribell, Cognizance Biomarkers, Cowen and Company, LLC, Crossject, EcoR1 Capital, Eisai, Encoded Therapeutics, Engrail, Epalex, EpiMinder, Epitel Inc., Equilibre BioPharmaceuticals, Genentech Inc., Grin Therapeutics, iQure Pharma Inc, IQVIA RDS Inc., Janssen Pharmaceutica, Jazz Pharmaceuticals, Korro Bio Inc., Leal Therapeutics Inc., Lipocine, LivaNova, Longboard Pharmaceuticals, Marinus, Modulight.bio, Neumirna Therapeutics, Neurelis, Neurocrine, NeuroPace Inc., NeuroPro Therapeutics, Neuroventis, Neurona Therapeutics, Ono Pharmaceutical Co., Otsuka Pharmaceutical Development, Ovid Therapeutics Inc., Praxis, PureTech LTY Inc., Rapport Therapeutics Inc., Receptor Holdings Inc., River0vest Venture Partners, Sage Therapeutics Inc., SK Life Sciences, Stoke, Supernus, Takeda, Taysha Gene Therapies, Third Rock Ventures LLC, UCB Inc., Ventus Therapeutics, Vida Ventures Management, and Xenon; and has received research support from the Epilepsy Study Consortium (funded by Eisai and UCB), the Epilepsy Study Consortium/Epilepsy Foundation, (funded by UCB), GW/FACES/One8Foundation, and the National Institute of Neurological Disorders and Stroke, outside the submitted work. All other authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.